Socialism and data science


Perhaps the best argument for capitalism is that no central planner can adequately understand the entire economy, however good their intentions. This point is made by Hayek, as Cass Sunstein summarizes in his book Infotopia:

Hayek claims that the great advantage of prices is that they aggregate both the information and the tastes of numerous people, incorporating far more material than could possibly be assembled by any central planner or board… For Hayek, the key economics question is how to incorporate that unorganized and dispersed knowledge.  That problem cannot possibly be solved by any particular person or board.  Central planners cannot have access to all of the knowledge held by particular people.  Taken as a whole, the knowledge held by those people is far greater than that held by even the most well-chosen experts.

Free market advocates are prone to take this point too far, overlooking pervasive market failures and the well documented merits of the welfare state and mixed economies. Nonetheless, it remains a central argument for the use of markets to organize much of our economy.

But what if planners could actually aggregate and make sense of all that information?

There’s an interesting piece in The New Yorker on the history of this idea, the interplay between “big data” and socialism. The piece looks at Chile under Allende, and the quest to utilize computer modeling to aid central planning:

At the center of Project Cybersyn (for “cybernetics synergy”) was the Operations Room, where cybernetically sound decisions about the economy were to be made. Those seated in the op room would review critical highlights—helpfully summarized with up and down arrows—from a real-time feed of factory data from around the country… Four screens could show hundreds of pictures and figures at the touch of a button, delivering historical and statistical information about production—the Datafeed… In addition to the Datafeed, there was a screen that simulated the future state of the Chilean economy under various conditions. Before you set prices, established production quotas, or shifted petroleum allocations, you could see how your decision would play out.

As you can imagine, the modeling that was possible in the 70’s wasn’t all that sophisticated, and so it’s no surprise that the system didn’t overcome Hayek’s critique. But the example is a reminder that the efficacy of central planning isn’t necessarily static. With increasingly ubiquitous data collection and more and more advanced data analysis tools and even artificial intelligence, might planners one day rival the market’s ability to distribute scarce resources?

We’re not nearly there yet, however numerous pieces in recent weeks have detailed government’s increasing interest in data science as a tool for conducting policy. Cities like Chicago are using analytics to improve public health, by better targeting regulators’ interventions based on predictive models. More ambitiously, India now has a dashboard that logs attendance of government workers throughout the country. Perhaps the most aggressive is Singapore, which is collecting a frankly scary amount of data:

Across Singapore’s national ministries and departments today, armies of civil servants use scenario-based planning and big-data analysis from RAHS for a host of applications beyond fending off bombs and bugs. They use it to plan procurement cycles and budgets, make economic forecasts, inform immigration policy, study housing markets, and develop education plans for Singaporean schoolchildren — and they are looking to analyze Facebook posts, Twitter messages, and other social media in an attempt to “gauge the nation’s mood” about everything from government social programs to the potential for civil unrest.

There are any number of objections to raise here, starting with privacy. And The New Yorker piece makes clear that the fragile political economy in Chile mattered more than modeling limitations in limiting that particular experiment. Public choice critiques of planning, based on bureaucratic incentives, likely remain poignant even as technical barriers are removed.

And yet in an age where we are able to imagine automated offices, robotic managers, and markets ruled in real-time by algorithms, why not allow ourselves to consider, even briefly, what the same technologies could do for government? Not just to improve a rule here or a program there, but to perhaps revise what the optimal economic system looks like.

My favorite, if oversimplified ,description of the choice between markets and central planning comes from political scientist Charles Lindblom. As I’ve written of it previously:

In his 1977 book “Politics and Markets”, political scientist Charles Lindblom describes the “key difference” between markets and central planning as “the role of intellect in social organization” with “on the one side, a confident distinctive view of man using his intelligence in social organization [central planning]; on the other side, a skeptical view of his capacity [markets].”

If this is true, then the belief that ubiquitous data collection, cheap computing power, and machine intelligence are making us smarter over time should be mirrored by a belief that planning is becoming more plausible. Perhaps more techo-utopians ought to be aspiring socialists, too.

Image via